Close

1. Identity statement
Reference TypeConference Paper (Conference Proceedings)
Sitemarte.sid.inpe.br
Holder Codeisadg {BR SPINPE} ibi 8JMKD3MGPCW/3DT298S
Identifier3ERPFQRTRW/39UFKJE
Repositorydpi.inpe.br/marte/2011/06.27.14.17
Last Update2011:06.27.14.17.51 (UTC) administrator
Metadata Repositorydpi.inpe.br/marte/2011/06.27.14.17.51
Metadata Last Update2018:06.06.02.23.32 (UTC) administrator
ISBN978-85-17-00056-0 (Internet)
978-85-17-00057-7 (DVD)
Citation KeyMedeirosCastErthDutr:2011:ClImMé
TitleClassificação de imagens pelo método de Árvore de Decisão Oblíqua
FormatDVD, Internet.
Year2011
Access Date2024, May 19
Secondary TypePRE CN
Number of Files1
Size282 KiB
2. Context
Author1 Medeiros, Ivo Paixão de
2 Castro Filho, Carlos Alberto Pires de
3 Erthal, Guaraci José
4 Dutra, Luciano Vieira
Resume Identifier1
2
3 8JMKD3MGP5W/3C9JHBN
4 8JMKD3MGP5W/3C9JHMA
Group1
2 DPI-OBT-INPE-MCT-BR
3 DPI-OBT-INPE-MCT-BR
4 DPI-OBT-INPE-MCT-BR
Affiliation1 Instituto Tecnológico da Aeronáutica - ITA
2 Instituto Nacional de Pesquisas Espaciais - INPE
3 Instituto Nacional de Pesquisas Espaciais - INPE
4 Instituto Nacional de Pesquisas Espaciais - INPE
Author e-Mail Address1 ivopdm@gmail.com
2 pires@dpi.inpe.br
3 gaia@dpi.inpe.br
4 dutra@dpi.inpe.br
EditorEpiphanio, José Carlos Neves
Galvão, Lênio Soares
e-Mail Addressluana@dsr.inpe.br
Conference NameSimpósio Brasileiro de Sensoriamento Remoto, 15 (SBSR).
Conference LocationCuritiba
Date30 abr. - 5 maio 2011
PublisherInstituto Nacional de Pesquisas Espaciais (INPE)
Publisher CitySão José dos Campos
Pages4255-4262
Book TitleAnais
OrganizationInstituto Nacional de Pesquisas Espaciais (INPE)
History (UTC)2011-08-15 15:17:44 :: luana@dsr.inpe.br -> administrator :: 2011
2018-06-06 02:23:32 :: administrator -> tereza@sid.inpe.br :: 2011
3. Content and structure
Is the master or a copy?is the master
Content Stagecompleted
Transferable1
Keywordspattern recognition
classification
decision tree
reconhecimento de padrões
AbstractThe Oblique Decision Tree (ODT) classification method has the advantage of dividing feature spaces using multidimensional hyperplanes that are oblique to the Cartesian axes. Because it is a non-parametric classification method, it also is capable of classifying images containing different statistical distribution. This paper aims to present a model of an ODT and perform tests comparing its classification results with other traditional classification methods. The ODT developed is binary and uses the Exchange Method for splitting each node into two subsets of classes, along with Fisher´s Linear Discriminant to calculate the parameters for the hyperplanes. Tests have been done using Polarimetric Interferometric and Synthetic Aperture Radar data from the brazilian Terrestrial Cartography Subproject, also known as Amazon Radiography, of the Geographic Service of Brazilian Army (DSG). Throughout the tests the ODT showed slightly better classification results compared to the Ordinary Binary Classification Tree, obtaining better overall accuracy (86.06%) and smaller tree. Besides that, the ODT showed better results than those obtained with traditional classifiers such as the Maximum Likelihood (85.44%), Nearest Neighbor (77.07%) and Mahalanobis Distance (78.72%). In the other hand, the Support Vector Machine classification method obtained higher overall accuracy (87.37%) although a much higher processing time is needed.
AreaSRE
TypeProcessamento de Imagens
doc Directory Contentaccess
source Directory Contentthere are no files
agreement Directory Contentthere are no files
4. Conditions of access and use
data URLhttp://urlib.net/ibi/3ERPFQRTRW/39UFKJE
zipped data URLhttp://urlib.net/zip/3ERPFQRTRW/39UFKJE
Languagept
Target Filep0953.pdf
User Groupadministrator
luana@dsr.inpe.br
Visibilityshown
5. Allied materials
Mirror Repositoryurlib.net/www/2011/03.29.20.55
Next Higher Units8JMKD3MGPCW/3EQCCU5
Citing Item Listsid.inpe.br/mtc-m21/2012/07.13.14.53.50 1
Host Collectiondpi.inpe.br/banon/2003/12.10.19.30
6. Notes
Empty Fieldsarchivingpolicy archivist callnumber contenttype copyholder copyright creatorhistory descriptionlevel dissemination documentstage doi edition issn label lineage mark nextedition notes numberofvolumes orcid parameterlist parentrepositories previousedition previouslowerunit progress project readergroup readpermission rightsholder schedulinginformation secondarydate secondarykey secondarymark serieseditor session shorttitle sponsor subject tertiarymark tertiarytype url versiontype volume
7. Description control
e-Mail (login)tereza@sid.inpe.br
update 


Close